The NBT spec requires that all strings are UTF-8 encoded. However in
practice, non-UTF-8 data can get into a level.dat file (generally via
some weird user-supplied characters in signs. This results in the
following error:
CorruptNBTError: could not parse nbt: 'utf8' codec can't decode byte
0xc0 in position 3: invalid start byte
This happens often enough that we should just ignore this error by using
the 'replace' strategy to replace the invalid data